]> git.donarmstrong.com Git - samtools.git/blob - sam.c
* samtools-0.1.3-14 (r262)
[samtools.git] / sam.c
1 #include <string.h>
2 #include "sam.h"
3
4 #define TYPE_BAM  1
5 #define TYPE_READ 2
6
7 bam_header_t *bam_header_dup(const bam_header_t *h0)
8 {
9         bam_header_t *h;
10         int i;
11         h = bam_header_init();
12         *h = *h0;
13         h->hash = 0;
14         h->text = (char*)calloc(h->l_text + 1, 1);
15         memcpy(h->text, h0->text, h->l_text);
16         h->target_len = (uint32_t*)calloc(h->n_targets, 4);
17         h->target_name = (char**)calloc(h->n_targets, sizeof(void*));
18         for (i = 0; i < h->n_targets; ++i) {
19                 h->target_len[i] = h0->target_len[i];
20                 h->target_name[i] = strdup(h0->target_name[i]);
21         }
22         return h;
23 }
24
25 samfile_t *samopen(const char *fn, const char *mode, const void *aux)
26 {
27         samfile_t *fp;
28         fp = (samfile_t*)calloc(1, sizeof(samfile_t));
29         if (mode[0] == 'r') { // read
30                 fp->type |= TYPE_READ;
31                 if (mode[1] == 'b') { // binary
32                         fp->type |= TYPE_BAM;
33                         fp->x.bam = strcmp(fn, "-")? bam_open(fn, "r") : bam_dopen(fileno(stdin), "r");
34                         if (fp->x.bam == 0) goto open_err_ret;
35                         fp->header = bam_header_read(fp->x.bam);
36                 } else { // text
37                         fp->x.tamr = sam_open(fn);
38                         if (fp->x.tamr == 0) goto open_err_ret;
39                         fp->header = sam_header_read(fp->x.tamr);
40                         if (fp->header->n_targets == 0) { // no @SQ fields
41                                 if (aux) { // check if aux is present
42                                         bam_header_destroy(fp->header);
43                                         fp->header = sam_header_read2((const char*)aux);
44                                 }
45                                 if (fp->header->n_targets == 0)
46                                         fprintf(stderr, "[samopen] empty header.\n");
47                         } else fprintf(stderr, "[samopen] SAM header is present: %d sequences.\n", fp->header->n_targets);
48                 }
49         } else if (mode[0] == 'w') { // write
50                 fp->header = bam_header_dup((const bam_header_t*)aux);
51                 if (mode[1] == 'b') { // binary
52                         fp->type |= TYPE_BAM;
53                         fp->x.bam = strcmp(fn, "-")? bam_open(fn, "w") : bam_dopen(fileno(stdout), "w");
54                         if (fp->x.bam == 0) goto open_err_ret;
55                         bam_header_write(fp->x.bam, fp->header);
56                 } else { // text
57                         // open file
58                         fp->x.tamw = strcmp(fn, "-")? fopen(fn, "w") : stdout;
59                         if (fp->x.tamr == 0) goto open_err_ret;
60                         // write header
61                         if (strstr(mode, "h")) {
62                                 int i;
63                                 bam_header_t *alt;
64                                 // parse the header text 
65                                 alt = bam_header_init();
66                                 alt->l_text = fp->header->l_text; alt->text = fp->header->text;
67                                 sam_header_parse(alt);
68                                 alt->l_text = 0; alt->text = 0;
69                                 // check if there are @SQ lines in the header
70                                 if (alt->n_targets) { // then write the header text without dumping ->target_{name,len}
71                                         if (alt->n_targets != fp->header->n_targets)
72                                                 fprintf(stderr, "[samopen] inconsistent number of target sequences.\n");
73                                         fwrite(fp->header->text, 1, fp->header->l_text, fp->x.tamw);
74                                 } else { // then dump ->target_{name,len}
75                                         for (i = 0; i < fp->header->n_targets; ++i)
76                                                 fprintf(fp->x.tamw, "@SQ\tSN:%s\tLN:%d\n", fp->header->target_name[i], fp->header->target_len[i]);
77                                 }
78                                 bam_header_destroy(alt);
79                         }
80                 }
81         }
82         return fp;
83
84 open_err_ret:
85         free(fp);
86         return 0;
87 }
88
89 void samclose(samfile_t *fp)
90 {
91         if (fp == 0) return;
92         if (fp->header) bam_header_destroy(fp->header);
93         if (fp->type & TYPE_BAM) bam_close(fp->x.bam);
94         else if (fp->type & TYPE_READ) sam_close(fp->x.tamr);
95         else fclose(fp->x.tamw);
96         free(fp);
97 }
98
99 int samread(samfile_t *fp, bam1_t *b)
100 {
101         if (fp == 0 || !(fp->type & TYPE_READ)) return -1; // not open for reading
102         if (fp->type & TYPE_BAM) return bam_read1(fp->x.bam, b);
103         else return sam_read1(fp->x.tamr, fp->header, b);
104 }
105
106 int samwrite(samfile_t *fp, const bam1_t *b)
107 {
108         if (fp == 0 || (fp->type & TYPE_READ)) return -1; // not open for writing
109         if (fp->type & TYPE_BAM) return bam_write1(fp->x.bam, b);
110         else {
111                 char *s = bam_format1(fp->header, b);
112                 int l = strlen(s);
113                 fputs(s, fp->x.tamw); fputc('\n', fp->x.tamw);
114                 free(s);
115                 return l + 1;
116         }
117 }
118
119 int sampileup(samfile_t *fp, int mask, bam_pileup_f func, void *func_data)
120 {
121         bam_plbuf_t *buf;
122         int ret;
123         bam1_t *b;
124         b = bam_init1();
125         buf = bam_plbuf_init(func, func_data);
126         bam_plbuf_set_mask(buf, mask);
127         while ((ret = samread(fp, b)) >= 0)
128                 bam_plbuf_push(b, buf);
129         bam_plbuf_push(0, buf);
130         bam_plbuf_destroy(buf);
131         bam_destroy1(b);
132         return 0;
133 }